Prerequisites
- Python 3.13+
- uv package manager
- An OpenAI-compatible API key (set
OPENAI_API_KEY)
Install
git clone https://github.com/smith-nathanh/rho-agent.git
cd rho-agent
uv sync
Start an interactive session
export OPENAI_API_KEY=sk-...
uv run rho-agent main
This starts a REPL with the default readonly profile. The agent can inspect files and run read-only shell commands, but cannot modify anything.
Start a development session
uv run rho-agent main --profile developer --working-dir ~/proj/myapp
The developer profile enables file editing, unrestricted shell access, and the full tool suite.
Run a one-shot task
Pass a prompt as a positional argument to run a single task and exit:
uv run rho-agent main "list all Python files that import asyncio"
Use a prompt template
Prompt files are markdown documents with YAML frontmatter for variables:
uv run rho-agent main --prompt examples/job-failure.md \
--var cluster=prod \
--var log_path=/mnt/logs/123
See Prompt Files for the full template format.
Connect to a database
Set the appropriate environment variables and the database tools become available automatically:
export SQLITE_DB=/path/to/data.db
uv run rho-agent main "list all tables and describe their schemas"
Database tools support PostgreSQL, MySQL, Oracle, Vertica, and SQLite. See Tools for configuration details.
Monitor running agents
# List running agents
uv run rho-agent ps
# Launch the observability dashboard
uv run rho-agent dashboard
# Open the interactive monitor
uv run rho-agent monitor
Next steps
- Installation — all install methods and environment configuration
- CLI Reference — complete command and flag documentation
- Profiles — understand and customize capability profiles
- Runtime API — embed agents in Python services